Introduction

As large multivariate datasets become increasingly common we require more effective ways to display, analyze, filter and interpret this large amount of information.

The concern for increasing size challenges fundamental methods that have been designed and conceptually verified on moderately sized datasets. This challenge manifests itself in methods across many fields, from computational complexity to database organizations to the visual presentation of data. The latter is the subject matter of this paper.

Our concern with visualization reaches beyond data display. It extends toward data exploration, in seeking and unfolding patterns not immediately obvious or comprehensible. It is hence an active process of discovery as opposed to passive display. And it is through data exploration that meaningful ideas, relations, and subsequent inferences are extracted from the data.

What is multivariate dataset ?

• An N-dimensional dataset E comprises elements e_i = (x_i1,x_i2,...,x_in).
• Each observation x_ij may be independent of or interdependent on one or more of the other observations.
• Observations may be discrete or continuous in nature, or may take on nominal values.

Examples of the types of multivariate data

• Physical interpretation such as geographical data
• A sequence of time-varying information such as stock prices